TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Reverse OCR

745 pointsby mrtbldover 10 years ago

28 comments

albertzeyerover 10 years ago
I was thinking about this: <a href="http://www.cs.toronto.edu/~graves/handwriting.html" rel="nofollow">http:&#x2F;&#x2F;www.cs.toronto.edu&#x2F;~graves&#x2F;handwriting.html</a>
评论 #8564642 未加载
评论 #8563091 未加载
评论 #8561891 未加载
评论 #8564220 未加载
评论 #8562041 未加载
praptakover 10 years ago
This is similar to the project where images of clouds were fed to face recognition software: <a href="http://ssbkyh.com/works/cloud_face/" rel="nofollow">http:&#x2F;&#x2F;ssbkyh.com&#x2F;works&#x2F;cloud_face&#x2F;</a>
评论 #8562515 未加载
评论 #8562279 未加载
jparishyover 10 years ago
Not strictly related, but reminded me of the exercise in genetic programming by Roger Alsing: <a href="http://rogeralsing.com/2008/12/07/genetic-programming-evolution-of-mona-lisa/" rel="nofollow">http:&#x2F;&#x2F;rogeralsing.com&#x2F;2008&#x2F;12&#x2F;07&#x2F;genetic-programming-evolut...</a><p>It&#x27;s a rather cool attempt to draw the Mona Lisa using random, semi-transparent polygons
评论 #8562646 未加载
baneover 10 years ago
This could be a cool way to visually &quot;encrypt&quot; messages. They&#x27;re readable, but only by the correct tool. I wonder how these squiggles might be creatively arranged steganographicly in an image and still be &quot;read&quot; by the OCR tool.
评论 #8562219 未加载
评论 #8562885 未加载
评论 #8562062 未加载
kitdover 10 years ago
Could be used for automated printing of doctors&#x27; prescriptions ;)
mrtbldover 10 years ago
Perhaps this could lead to a new kind of captcha that only bots can solve. I doubt it would be efficient, though.
评论 #8562178 未加载
评论 #8562591 未加载
评论 #8562320 未加载
评论 #8562138 未加载
评论 #8562093 未加载
评论 #8562052 未加载
carsonreinkeover 10 years ago
Looks like he has written tons of very creative bots. They are all very interesting ideas (e.g. <a href="http://randomshopper.tumblr.com" rel="nofollow">http:&#x2F;&#x2F;randomshopper.tumblr.com</a>)
评论 #8564629 未加载
sgentleover 10 years ago
It would be pretty interesting to see one degree of abstraction up from this - what sets of lines are close enough to match a certain word?<p>If you averaged over all those sets, would the resulting blobby heatmap resemble the original word in a legible form? Or something else?
userbinatorover 10 years ago
I can imagine generating a few pages or even an entire book of this, and some future generations attempting to figure out what sort of language it was written in... reminds me of this:<p><a href="http://en.wikipedia.org/wiki/Voynich_manuscript" rel="nofollow">http:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Voynich_manuscript</a>
cosarara97over 10 years ago
I couldn&#x27;t get that OCR to read my mouse-written E. It&#x27;s a nice experiment nevertheless.
评论 #8561902 未加载
klausaover 10 years ago
I highly recommend watching talk Darius Kazemi (author of Reverse OCR) gave at this years XOXO: <a href="http://www.youtube.com/watch?v=l_F9jxsfGCw" rel="nofollow">http:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=l_F9jxsfGCw</a>
emhartover 10 years ago
It has been fantastic watching Darius&#x27; myriad experiments over the past few years. His work always has a great mixture of whimsy and serious experimentation.
MrBraover 10 years ago
Nice. Finally computers approached the age of writing. :)
lucb1eover 10 years ago
I can already imagine the innovation:<p>&gt; Type over this text to prove that you are a computer.<p>&gt; Human detected. Shoo, shoo!
Aaronneyerover 10 years ago
Looks like my handwriting
z3t4over 10 years ago
I can&#x27;t believe OCR has not been solved yet. The only one even close is OmniPage.
评论 #8566653 未加载
driverdanover 10 years ago
Here&#x27;s the source code on github: <a href="https://github.com/dariusk/reverseocr" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;dariusk&#x2F;reverseocr</a>
jostmeyover 10 years ago
A generative model, although computationally expensive, would not suffer this problem. Essentially a generative model can run in reverse, which means that if you feed values into the output you get inputs that could explain the output. Check out &quot;Boltzmann Machines&quot; for an example. There are plenty of examples for the MNIST dataset of hand written digits.
k_szeover 10 years ago
I think one of the problems is that the OCR assumes the images to be (English) letters.<p>To be really really useful, the OCR would need to consider at least all characters in the Unicode Basic Multilingual Plane. And then it needs to be able to reject an image as containing any word, and then it needs to solve the halting problem.
zwassover 10 years ago
This reminds me of an experiment I played with using random search to &quot;teach&quot; the browser how to draw characters: <a href="http://zwass.github.io/Learn2Write/" rel="nofollow">http:&#x2F;&#x2F;zwass.github.io&#x2F;Learn2Write&#x2F;</a>
bmh100over 10 years ago
This actually seems like a great program for automatically generating adversarial examples to improve OCR. A human could rate this text as being illegible or legible. Each example can then be added to the training data to improve its quality.
评论 #8565073 未加载
eurleifover 10 years ago
It would be neat to see the same thing, except using two OCR libraries instead of just one, and requiring both libraries to be able to read the message. I imagine the letters would start to look a bit less insane.
shangxiaoover 10 years ago
This is pretty cool, although it makes me wonder what the real world applications could be. It does, at the very least, tantalise my curiosity and gets me thinking.
achr2over 10 years ago
Could this be used in a pseudo reverse CAPTCHA by showing a series of words, and asking the user to say which is not human readable?
methylover 10 years ago
I wonder what would happen if you run this program letter-by-letter, possibly the readability could increase.
mslotover 10 years ago
I love algorithmic art.
Applicoover 10 years ago
very cool idea.
jdimovover 10 years ago
What (if anything) is this saying about the quality of the OCR process? Especially since none of these seem human readable.
评论 #8562251 未加载
评论 #8562326 未加载
评论 #8563325 未加载
评论 #8563329 未加载